Вы находитесь на странице: 1из 19

ASSESSMENT 10.

1177/1073191105283504
Baer et al. / FACETS OF MINDFULNESS

Using Self-Report Assessment Methods


to Explore Facets of Mindfulness

Ruth A. Baer
Gregory T. Smith
Jaclyn Hopkins
Jennifer Krietemeyer
Leslie Toney
University of Kentucky

The authors examine the facet structure of mindfulness using five recently developed mind-
fulness questionnaires. Two large samples of undergraduate students completed mindful-
ness questionnaires and measures of other constructs. Psychometric properties of the
mindfulness questionnaires were examined, including internal consistency and convergent
and discriminant relationships with other variables. Factor analyses of the combined pool of
items from the mindfulness questionnaires suggested that collectively they contain five clear,
interpretable facets of mindfulness. Hierarchical confirmatory factor analyses suggested
that at least four of the identified factors are components of an overall mindfulness construct
and that the factor structure of mindfulness may vary with meditation experience. Mindful-
ness facets were shown to be differentially correlated in expected ways with several other
constructs and to have incremental validity in the prediction of psychological symptoms.
Findings suggest that conceptualizing mindfulness as a multifaceted construct is helpful in
understanding its components and its relationships with other variables.

Keywords: mindfulness; questionnaires; self-report assessment; factor structure; facets; meditation

Mindfulness is usually defined to include bringing 1982, 1990), mindfulness-based cognitive therapy (MBCT;
one’s complete attention to the experiences occurring in Segal, Williams, & Teasdale, 2002), acceptance and com-
the present moment, in a nonjudgmental or accepting way mitment therapy (ACT; S. C. Hayes, Strosahl, & Wilson,
(Brown & Ryan, 2003; Kabat-Zinn, 1990; Linehan, 1999), and relapse prevention for substance abuse (Marlatt
1993a; Marlatt & Kristeller, 1999). Descriptions of mind- & Gordon, 1985; Parks, Anderson, & Marlatt, 2001) as
fulness and methods for cultivating it originate in eastern well as variations on these approaches. These interven-
spiritual traditions, which suggest that mindfulness can be tions conceptualize mindfulness as a set of skills that can
developed through the regular practice of meditation, and be learned and practiced in order to reduce psychological
that increases in positive qualities such as awareness, in- symptoms and increase health and well-being. MBSR and
sight, wisdom, compassion, and equanimity are likely to MBCT rely heavily on formal meditation practices, in
result (Goldstein, 2002; Kabat-Zinn, 2000). In recent de- which participants spend up to 45 minutes each day direct-
cades, traditional mindfulness meditation practices have ing their attention in specific ways. In contrast, DBT and
been adapted for secular use and incorporated into several ACT rely on a wide variety of shorter exercises in which
interventions that are now widely available in medical and mindfulness-related skills can be practiced without neces-
mental health settings. These interventions include dia- sarily engaging in meditation.
lectical behavior therapy (DBT; Linehan, 1993a, 1993b), The empirical literature increasingly supports the effi-
mindfulness-based stress reduction (MBSR; Kabat-Zinn, cacy of mindfulness-based interventions. Reductions in

Assessment, Volume 13, No. 1, March 2006 27-45


DOI: 10.1177/1073191105283504
© 2006 Sage Publications
28 ASSESSMENT

symptoms have been reported across a wide range of pop- factor described as attention to and awareness of what is
ulations and disorders (Baer, 2003; S. C. Hayes, Masuda, taking place in the present. They argued that acceptance is
Bissett, Luoma, & Guerrero, 2004; Robins & Chapman, important to mindfulness but that it is subsumed within the
2004). Until very recently, however, the assessment of capacity to pay full attention to the present moment.
mindfulness has received much less empirical attention. The primary purpose of the project described here was
Dimidjian and Linehan (2003a) noted that psycho- to examine the facet structure of the mindfulness con-
metrically sound measures of mindfulness are necessary struct, because it has been operationalized in several inde-
for understanding the nature of mindfulness and its com- pendently developed self-report questionnaires. First, Part
ponents and the mechanisms by which mindfulness train- 1 examined the psychometric characteristics of the avail-
ing exerts its beneficial effects. Brown and Ryan (2004) able mindfulness questionnaires, including internal con-
and Bishop et al. (2004) made similar points, arguing that sistency, correlations with each other, and convergent and
operational definitions of mindfulness are essential for the discriminant relationships with a variety of other con-
development of valid instruments, which in turn are neces- structs. With the psychometric soundness of these ques-
sary for investigating the psychological processes in- tionnaires reasonably well established, Part 2 then com-
volved in mindfulness training. bined all items from the available questionnaires into a
Within the past few years, self-report questionnaires single data set and used exploratory factor analysis to ex-
for the assessment of mindfulness have begun to appear in amine the facet structure of this combined item pool. A
the literature. The development of these questionnaires is five-facet structure was derived. Part 3 used confirmatory
an important advance in the study of mindfulness because factor analysis to examine the validity of this facet struc-
it provides new opportunities for empirical investigations ture in an independent sample. In Part 4, differential rela-
of the nature of mindfulness and its relationships with tionships between identified facets and several measures
other psychological constructs. As the process of writing of conceptually related constructs were explored. In Part
items for any self-report questionnaire requires authors to 5, incremental validity of mindfulness facets in the predic-
define or conceptualize the construct they are attempting tion of psychological symptoms was examined. Methods
to measure (Clark & Watson, 1995), each available mind- and findings are described following a brief overview of
fulness questionnaire represents an attempt to operation- the available mindfulness questionnaires.
alize mindfulness by writing self-report items that capture
its essence. Empirical examination of these questionnaires Available Mindfulness Questionnaires
could provide important information about how mindful-
ness should be defined and described. The Mindful Attention Awareness Scale (MAAS;
Among the important questions that can be studied us- Brown & Ryan, 2003) is a 15-item instrument measuring
ing these instruments is whether mindfulness should be the general tendency to be attentive to and aware of
described as a multifaceted construct and, if so, how the present-moment experience in daily life. It has a single-
facets should be defined. Several current descriptions of factor structure and yields a single total score. Using a 6-
mindfulness suggest a multidimensional nature. For ex- point Likert-type scale (almost always to almost never),
ample, in DBT (Dimidjian & Linehan, 2003b) mind- respondents rate how often they have experiences of act-
fulness is conceptualized as having six elements: three ing on automatic pilot, being preoccupied, and not paying
related to what one does when being mindful (observing, attention to the present moment. Items include, “I find my-
describing, and participating) and three related to how one self doing things without paying attention,” and “I break or
does it (nonjudgmentally, one-mindfully, and effectively). spill things because of carelessness, not paying attention,
Segal et al. (2002) summarized the nature of mindfulness or thinking of something else.” The authors reported inter-
by stating, nal consistency (coefficient alpha) of .82 and expected
convergent and discriminant validity correlations. For ex-
In mindfulness practice, the focus of a person’s at- ample, the MAAS was significantly positively correlated
tention is opened to admit whatever enters experi- with openness to experience, emotional intelligence, and
ence, while at the same time, a stance of kindly well-being; negatively correlated with rumination and so-
curiosity allows the person to investigate whatever cial anxiety; and unrelated to self-monitoring. MAAS
appears, without falling prey to automatic judg-
scores also were significantly higher in mindfulness prac-
ments or reactivity. (pp. 322-323)
titioners than in matched community controls. In a group
This description suggests several elements, including ob- of cancer patients who completed an MBSR course, in-
servation of present-moment experience, acceptance, non- creases in MAAS scores were associated with decreases in
judging, and nonreactivity. On the other hand, Brown and mood disturbance and symptoms of stress.
Ryan (2004) argued that mindfulness consists of a single The Freiburg Mindfulness Inventory (FMI; Buchheld,
Grossman, & Walach, 2001) is a 30-item instrument as-
Baer et al. / FACETS OF MINDFULNESS 29

sessing nonjudgmental present-moment observation and ings I have.” The authors reported internal consistencies of
openness to negative experience. It was developed with .74 to .80; negative correlations with experiential avoid-
participants in mindfulness meditation retreats and is de- ance, thought suppression, rumination, worry, depression,
signed for use with experienced meditators. Items are and anxiety; and positive correlations with clarity of feel-
rated on a 4-point Likert-type scale (rarely to almost al- ings, mood repair, cognitive flexibility, and well-being
ways). Items include, “I watch my feelings without be- (Feldman et al., 2004; S. C. Hayes & Feldman, 2004). In-
coming lost in them,” and “I am open to the experience of creases in mindfulness scores were observed in a sample
the present moment.” The authors reported internal con- of individuals completing an integrative therapy for de-
sistencies of .93 and .94 in individuals who completed the pression that includes a mindfulness component (A. M.
inventory at the beginning and end of intensive meditation Hayes & Harris, 2000).
retreats lasting from 3 to 14 days. Mean score increased by The Mindfulness Questionnaire (MQ; Chadwick,
about 1 standard deviation from preretreat to postretreat. Hember, Mead, Lilley, & Dagnan, 2005) is a 16-item in-
Correlations with measures of other constructs were not strument assessing a mindful approach to distressing
reported. Although exploratory factor analyses suggested thoughts and images. All items begin with, “Usually, when
a four factor solution, the solution was somewhat unstable I have distressing thoughts or images” and continue with a
from preretreat to postretreat, and many items loaded on mindfulness-related response, such as, “I am able just to
more than one factor. The authors suggested that the scale notice them without reacting” and “I am able to accept the
should be interpreted unidimensionally and recommend experience.” Items are rated on a 7-point Likert-type scale
use of a single total score. (agree totally to disagree totally). The authors noted that
The Kentucky Inventory of Mindfulness Skills (KIMS; items represent four aspects of mindfulness: mindful ob-
Baer, Smith, & Allen, 2004) is a 39-item instrument de- servation, letting go, nonaversion, and nonjudgment, but
signed to measure four elements of mindfulness: observ- that a unidimensional factor structure provided the best fit
ing, describing, acting with awareness, and accepting to their data. Thus, the computation of subscale scores is
without judgment. Items include, “I notice when my not recommended. The authors reported good internal
moods begin to change” (observe); “I’m good at finding consistency (alpha = .89), a significant correlation with the
words to describe my feelings” (describe); “When I do MAAS (r = .57), significant differences in the expected di-
things, my mind wanders off and I’m easily distracted” rection between meditators and nonmeditators, a signifi-
(act with awareness); and “I tell myself that I shouldn’t be cant positive correlation with mood ratings, and a signifi-
feeling the way I’m feeling” (accept without judgment). cant increase in scores for participants in an MBSR course.
Items are rated on a 5-point Likert-type scale (never or
very rarely true to always or almost always true). The
KIMS is based largely on the DBT conceptualization of PART 1: PSYCHOMETRIC CHARACTERISTICS
mindfulness skills. It measures a general tendency to be OF MINDFULNESS QUESTIONNAIRES
mindful in daily life and does not require experience with
meditation. Internal consistencies range from .76 to .91 for The purpose of Part 1 was to examine whether the avail-
the four subscales. Exploratory and confirmatory factor able mindfulness questionnaires are internally consistent
analyses clearly support the proposed four-factor struc- and correlated with each other, with meditation experi-
ture, and expected correlations with a variety of other con- ence, and with measures of other constructs expected to be
structs were obtained. Scores were found to be signifi- related or unrelated to mindfulness.
cantly lower in a sample of individuals with borderline
personality disorder than in a student sample for three of Method
the four scales (Baer et al., 2004).
The Cognitive and Affective Mindfulness Scale (CAMS; Participants
Feldman, Hayes, Kumar, & Greeson, 2004; S. C. Hayes &
Feldman, 2004) is a 12-item inventory designed to mea- Participants were 613 undergraduate psychology stu-
sure attention, awareness, present-focus, and acceptance/ dents (Sample 1) who completed the procedures for credit
nonjudgment with respect to thoughts and feelings in in their classes. Their mean age was 20.5 years (range =
general daily experience. Although it attempts to capture 18-57), 70% were female, and 90% were Caucasian.
several elements of mindfulness, it does not measure them
separately but yields a single total score. Items are rated on Procedures
a 4-point Likert-type scale (rarely/not at all to almost al-
ways). Items include, “I try to notice my thoughts without Sessions were conducted with 20 to 25 students and
judging them,” “It is easy for me to concentrate on what I lasted about 60 minutes. After signing consent forms, par-
am doing,” and “I am able to accept the thoughts and feel- ticipants completed a packet of questionnaires, beginning
30 ASSESSMENT

with a short demographic form, which requested their age, depression and more life satisfaction (Martinez-Pons,
gender, year in school, race, and experience with medita- 1997). Because many descriptions of mindfulness include
tion. The latter variable was rated on a 5-point scale rang- observation and description of feelings, positive corre-
ing from 1 (none) to 5 (a lot). For all participants, the lations between TMMS and mindfulness scores were pre-
packet also included the five mindfulness questionnaires dicted.
described earlier (MAAS, FMI, KIMS, CAMS, MQ). The
White Bear Suppression Inventory (WBSI; Wegner &
FMI was included in spite of its authors’ concerns that
Zanakos, 1994). The WBSI measures thought suppres-
items’ meanings may not be clear to individuals without
sion, or deliberate attempts to avoid or get rid of unwanted
meditation experience. Close inspection of the items sug-
thoughts. Paradoxically, such attempts have been found to
gested that nonmeditators may be able to respond to them
increase the frequency of these thoughts (Wenzlaff &
meaningfully. Because its developers did not test the FMI
Wegner, 2000). Because mindfulness includes acceptance
with nonmeditators, its characteristics in this group are un-
of all thoughts as they occur and allowing them to come
known. Each participant also completed several measures
and go, negative correlations between WBSI scores and
of psychological constructs predicted to be related or unre-
mindfulness scales were predicted.
lated to mindfulness (described later). After the demo-
graphic form, the order of all instruments was randomized. Difficulties in Emotion Regulation Scale (DERS; Gratz
& Roemer, 2004). The DERS measures several elements
Measures and Predictions of emotion regulation, including awareness, understand-
ing, and acceptance of emotions as well as ability to act in
Each participant completed a subset of the following desired ways regardless of emotional state and access to
measures. Time constraints did not allow all participants to emotion regulation strategies. Gratz and Roemer (2004)
complete all measures. reported internal consistency of .93, test-retest reliability
of .88 during a 4- to 8-week interval, and a clear factor
Brief Symptom Inventory (BSI; Derogatis, 1992). The
structure and predicted significant correlations with sev-
BSI includes 53 items and provides scores for nine psy-
eral criterion variables, including experiential avoidance
chological symptom scales and a general severity index
and self-harm. Higher scores on the DERS indicate greater
(GSI). Only the GSI is reported here. Because the empiri-
difficulties in emotion regulation. Because mindfulness
cal literature shows that mindfulness practice is associated
includes awareness and acceptance of emotions, negative
with reduced symptoms, negative correlations between
correlations between DERS scores and mindfulness scales
the BSI and mindfulness scales were predicted.
were predicted.
NEO-Five Factor Inventory (NEO-FFI; Costa & Mc-
Toronto Alexithymia Scale (TAS-20; Bagby, Taylor, &
Crae, 1992). The NEO-FFI is a 60-item measure of the do-
Parker, 1993). Alexithymia includes difficulty identifying
mains of the five-factor model of personality, including
and describing feelings and a lack of interest in internal
neuroticism, extraversion, openness to experience, agree-
experience. The TAS-20 has shown good psychometric
ableness, and conscientiousness. Predictions were made
properties in student and clinical samples. Because mind-
for neuroticism, extraversion, and openness. Because
fulness includes interest in and observation of feelings,
mindfulness is associated with reduced negative affect,
negative correlations between TAS-20 and mindfulness
negative correlations between neuroticism and mindful-
scales were predicted.
ness scales were predicted. Many descriptions of mindful-
ness include attentiveness and receptivity to inner feelings Scale of Dissociative Activities (SODAS; Mayer &
and observation with interest of environmental stimuli, Farmer, 2003). The SODAS is a recently developed mea-
which seems consistent with the openness domain of the sure whose content includes acting without awareness,
five-factor model. Therefore, positive correlations be- lack of perception of inner experience, memory disrup-
tween Openness and Mindfulness scales were predicted. tions, and perceptions of unreality. It has shown good in-
Last, because level of mindfulness appears to be unrelated ternal consistency (alpha = .95) and test-retest stability
to introversion or extraversion, nonsignificant correlations during a 38-day interval (r = .77) and significant positive
were predicted. correlations with other measures of dissociation, includ-
ing experience sampling measures in naturalistic environ-
Trait Meta-Mood Scale (TMMS; Salovey, Mayer, Gold-
ments. Because mindfulness includes awareness of one’s
man, Turvey, & Palfai, 1995). The TMMS measures
inner experiences and actions, negative correlations be-
emotional intelligence, including attention to and clarity
tween the SODAS and mindfulness scales were predicted.
of feelings and ability to regulate feelings. Salovey et al.
(1995) have shown adequate to good internal consistency Acceptance and Action Questionnaire (AAQ; S. C. Hayes
for the TMMS, and higher scores are associated with less et al., in press). The AAQ measures experiential avoid-
Baer et al. / FACETS OF MINDFULNESS 31

ance, which is defined as negative evaluation of and un- TABLE 1


willingness to maintain contact with internal experiences, Intercorrelations Among Mindfulness Question-
such as bodily sensations, cognitions, emotions, and naires in Sample 1 (N = 613)
urges, and efforts to avoid, escape, or terminate these ex- FMI KIMS CAMS MQ
periences, even when doing so is harmful. It is associated
with increased levels of psychopathology and decreased MAAS .31** .51** .51** .38**
FMI — .57** .60** .45**
quality of life (S. C. Hayes, Wilson, Gifford, Follette, &
KIMS — .67** .45**
Strosahl, 1996). Because mindfulness includes observa- CAMS — .55**
tion and nonjudgmental acceptance of internal experi-
ences, negative correlations between AAQ scores and NOTE: MAAS = Mindful Attention Awareness Scale; FMI = Freiburg
Mindfulness Inventory; KIMS = Kentucky Inventory of Mindfulness
mindfulness scales were predicted. Skills; CAMS = Cognitive Affective Mindfulness Scale; MQ = Mindful-
ness Questionnaire.
Cognitive Failures Questionnaire (CFQ; Broadbent, **p < .01.
Cooper, Fitzgerald, & Parks, 1982). The CFQ assesses ab-
sent-mindedness, or the tendency to make errors on simple
tasks due to inattention, such as forgetting what to buy in tionships between mindfulness and other constructs.
the store or where one placed one’s keys. It has good inter- Facets of mindfulness are examined in later sections.
nal consistency (alpha =.89) and test-retest stability (.80-
.82) and is moderately correlated with boredom prone-
Internal Consistency and Intercorrelations
ness, attention deficit hyperactivity disorder in adults
(Wallace, Kass, & Stanny, 2002), and symptoms of stress,
The following alpha coefficients were obtained for the
anxiety, and depression but unrelated to intelligence, edu-
five mindfulness questionnaires, suggesting good inter-
cational level, or social desirability. Because mindfulness
nal consistency: MAAS = .86, FMI = .84, KIMS = .87,
should help individuals to avoid errors related to absent-
CAMS = .81, MQ = .85 (ns = 595-613). Relations among
mindedness, negative correlations between the CFQ and
the mindfulness questionnaires can be seen in Table 1. All
mindfulness scales were predicted.
are significantly positively correlated with each other,
Self-Compassion Scale (SCS; Neff, 2003a). Although with rs ranging from .31 (MAAS with FMI) to .67 (KIMS
self-compassion is a central element of the Buddhist psy- with CAMS).
chology from which mindfulness originates, efforts to
study it scientifically have emerged only recently. Neff Relationships With Meditation Experience
(2003b) suggested that self-compassion consists of sev-
eral elements, including a kind and nonjudgmental atti- Most of the sample had little or no meditation experi-
tude toward oneself when suffering; recognition that one’s ence, with 72% reporting none and 20% reporting a little.
experiences are part of the larger, more universal human To create a subsample with a somewhat more balanced
experience; and the holding of painful thoughts and feel- representation of meditation experience, we randomly
ings in balanced awareness, in which they are observed selected 20 participants who reported “none” and 20 who
and accepted without judgment, rumination, or self-pity. reported “a little” and combined them with the 42 who
Neff (2003a) conceptualized self-compassion as distinct reported “a medium amount” and the 6 who reported
from self-esteem in that it is nonevaluative. The SCS has “quite a bit” or “a lot,” for a subsample size of 88. For this
shown internal consistency of .92; test-retest reliability of subsample, correlations between meditation experience
.93 during a 3-week interval; significant positive correla- and mindfulness scales can be seen in the first line of Table
tions with social connectedness, emotional intelligence, 2. Correlations were significant and positive for the FMI
and life satisfaction; and significant negative correlations and KIMS. Correlations for the CAMS and MQ were
with self-criticism, perfectionism, depression, and anxi- nearly significant (ps = .06 and .07, respectively), sug-
ety. Because mindfulness includes awareness and accep- gesting that in a sample with better representation of medi-
tance of all experiences, with an attitude of acceptance and tation experience, these relationships might also be
nonjudging, positive correlations between SCS scores and significant.
mindfulness scales were predicted.
Mindfulness Measures and Other Variables:
Results and Discussion Convergent and Discriminant Correlations

The following analyses used total scores for all mea- Correlations between mindfulness measures and other
sures, because their purpose was to examine global rela- constructs are shown in Table 2. All correlations were in
32 ASSESSMENT

TABLE 2
Correlations Between Mindfulness Questionnaires and Other Variables
Mindfulness Questionnaire
Variable MAAS FMI KIMS CAMS MQ

Predicted positive correlations


a
Meditation experience –.04 .31** .33** .20 .20
Openness to experience .23* .30* .47** .22* .14
Emotional intelligence .22* .54** .61** .50** .27**
Self-compassion .36** .53** .49** .59** .57**
Predicted negative correlations
Psychological symptoms –.41** –.31** –.42** –.55** –.36**
Neuroticism –.41** –.53** –.37** –.63** –.58**
Thought suppression –.32** –.27** –.42** –.44** –.47**
Difficulties in emotion regulation –.34** –.46** –.56** –.63** –.58**
Alexithymia –.24** –.42** –.61** –.52** –.20**
Dissociation –.53** –.30* –.41** –.52** –.32*
Experiential avoidance –.32** –.54** –.44** –.51** –.60**
Absent-mindedness –.54** –.23** –.37** –.42** –.41**
Predicted nonsignificant correlations
Extraversion –.08 .20 .06 .05 .12

NOTE: MAAS = Mindful Attention Awareness Scale; FMI = Freiburg Mindfulness Inventory; KIMS = Kentucky Inventory of Mindfulness Skills;
CAMS = Cognitive Affective Mindfulness Scale; MQ = Mindfulness Questionnaire.
a. Selected subsample of n = 88.
*p <.05. **p < .01.

the expected directions, and all but one (MQ with open- facets of mindfulness. For example, the MAAS appears to
ness to experience) were statistically significant. Most emphasize an element of mindfulness related to dissocia-
were moderate to large (except those predicted to be tion and absent-mindedness, whereas the MQ focuses pri-
nonsignificant). These findings indicate that all of the marily on elements related to thought suppression and ex-
mindfulness scales show predicted relationships with periential avoidance.
other variables. Several authors have argued that measurement of com-
plex constructs at the facet level is important for clarifying
relationships between these constructs and other variables
PART 2: EXPLORING FACETS (Hough & Schneider, 1995; Paunonen & Ashton, 2001;
OF MINDFULNESS Schneider, Hough, & Dunnette, 1996; Smith, Fischer, &
Fister, 2003; Smith & McCarthy, 1995). Use of a single to-
Part 1 established that the available mindfulness ques- tal score for the target construct can obscure these relation-
tionnaires are internally consistent and, to a large extent, ships if facets of the target construct are differentially cor-
correlated with each other, with meditation experience, related with the other variables. That is, one or more facets
and in predictable directions with several other variables. of the target construct may be strongly related to a particu-
The findings also suggest that an examination of facets or lar variable, whereas other facets are not. Using a total
components of mindfulness might yield useful informa- score to examine these relationships effectively averages
tion about the nature of mindfulness and its relationships correlated with uncorrelated facets, providing a distorted
with other constructs. Close examination of Table 2 shows view of the relationship. An example of this problem can
that correlations between mindfulness and other variables be seen in the literature on the relationship between im-
vary widely. For example, emotional intelligence is corre- pulsivity and binge eating. Until recently, this literature
lated with the KIMS at .61 but with the MAAS at .22. A has found inconsistent relationships between these two
similar pattern can be seen with alexithymia. Absent- variables. However, more recent studies (Fischer, Smith &
mindedness is correlated with the MAAS at –.54 but with Anderson, 2003; Fischer, Smith, Anderson, & Flory, 2003)
the FMI at –.23, and experiential avoidance is correlated examining impulsivity at the facet level have shown that
with the MAAS at –.32 but with the MQ at –.60. (Differ- one facet of impulsivity known as urgency (tendency to act
ences between correlations noted here are significant at rashly when distressed) is strongly correlated with binge
p < .001). Such differences suggest that these question- eating, whereas other facets of impulsivity (sensation
naires may be measuring somewhat different elements or seeking, lack of premeditation) are not. Measures of im-
Baer et al. / FACETS OF MINDFULNESS 33

pulsivity that include some or all of these facets but do not mindfulness and should yield a broad-based empirical
assess them separately tend to provide ambiguous or dis- analysis of current thinking about the elements of mind-
torted information about this relationship. Similarly, it fulness. To the extent that this analysis yields factors con-
seems likely that clarification and reliable measurement of sistent with the KIMS, the validity of the KIMS will be
the facets of mindfulness might provide analogous ad- supported. However, the emergence of new facets might
vances in our understanding of this important construct. suggest ways to expand the assessment of mindfulness to
Analysis at the facet level is also important for examin- include more of the relevant content.
ing incremental validity in the assessment of mindfulness.
According to Haynes and Lench (2003), a measure has in- Method
cremental validity to the extent that it increases ability to
predict other measures of interest. Although incremental Participants and Procedures
validity can be examined at the level of individual instru-
ments (i.e., one could examine the incremental validity of Participants were the 613 students from Part 1. Their re-
existing mindfulness questionnaires in predicting other sponses to the items from the five mindfulness question-
measures), this procedure has disadvantages. It is unclear naires were combined into a single data set and subjected
what facets of mindfulness are included in some of the to the exploratory factor analyses and correlational analy-
available questionnaires, because they do not provide sub- ses described in the following sections.
scales. For this reason, evidence of incremental validity of
one over another may be hard to interpret. Smith et al. Results and Discussion
(2003) argued that reliable assessment of clearly specified
facets within a test is the most informative way of evalu- Exploratory Factor Analyses
ating incremental validity. That is, if facet scores can be
entered separately into regression analyses, then facets The combined data set for the five mindfulness ques-
significantly related to the dependent variable will be in- tionnaires (MAAS, FMI, KIMS, CAMS, MQ) included
cluded in the equation, whereas nonpredictive facets will 112 items. Using Sample 1 (n = 613), responses to this
be dropped, and the incremental validity of some facets combined item pool were subjected to exploratory factor
over others in predicting the dependent variable can be analysis (EFA) using principal axis factoring with oblique
examined. rotation to allow for correlations among the factors. Re-
This reasoning suggests that the most useful measures sults of the initial EFA yielded 26 factors with eigenvalues
of mindfulness will be those that measure all relevant fac- greater than 1.0 and accounting for 63% of the total vari-
ets separately and reliably. Among the mindfulness ques- ance. However, the scree plot clearly suggested a five-
tionnaires described earlier, only the KIMS provides factor solution. Floyd and Widamon (1995) argued that
subscales based on an empirically supported factor struc- the scree plot is a more useful guide to the number of fac-
ture. The MAAS has been shown to be unidimensional tors to retain, because use of eigenvalues greater than 1.0
(Brown & Ryan, 2003). The CAMS, FMI, and MQ con- can lead to overestimation of the number of meaning-
tain content pertaining to more than one facet but provide ful factors. Therefore, a second factor analysis was con-
only total scores, not subscale scores. Overall, it is unclear ducted, specifying that five factors should be identified
what facets of mindfulness may be represented in some of and, again, using principal axis factoring with oblique
these mindfulness questionnaires. rotation. This analysis yielded a five-factor solution ac-
The KIMS provides a clear facet structure with good counting for 33% of the variance after factor extraction.
empirical support (Baer et al., 2004). Given the strong cor- This factor structure is shown in Table 3. Only items with
relations between the KIMS and the other mindfulness minimum loadings of .40 on one factor and with a differ-
measures shown in Table 1, it seems likely that many of the ence of at least .20 between the highest and next highest
items from the other questionnaires measure facets similar factor loadings are included in the table. For each item, the
to those of the KIMS. However, it is also possible that questionnaire from which it originates and its item number
some items in the other questionnaires represent facets of on that questionnaire also are shown.
mindfulness not included in the KIMS. Therefore, the pur- Table 3 shows that four of the five factors are virtually
pose of Part 2 was to examine the facet structure of the identical to those identified in the development of the
combined item pool (112 items) from all five of the mind- KIMS (Baer et al., 2004) and that most of the KIMS items
fulness questionnaires from Part 1. Combining all items load on these factors, as do many items from the other
from these questionnaires into a single data set provides measures. An additional factor also emerged, with items
the opportunity to examine the facet structure across all of from the FMI and MQ that appear to describe a non-
these independently developed operationalizations of reactive stance toward internal experience.
34
TABLE 3
Factor Structure of Combined Items From Five Mindfulness Questionnaires in Sample of 613 Students
Factor Loading
Source of Item and Content 1 2 3 4 5

Factor 1: Nonreactivity to Inner Experience


*FMI 18: I perceive my feelings and emotions without having to react to them. .46 .14 .01 .00 .10
*FMI 25: I watch my feelings without getting lost in them. .44 .10 –.06 –.10 .07
*FMI 26: In difficult situations, I can pause without immediately reacting. .43 .18 –.12 .02 .05
*MQ 1: Usually when I have distressing thoughts or images, I am able just to notice them without reacting. .45 –.03 –.11 .05 .08
*MQ 4: Usually when I have distressing thoughts or images, I feel calm soon after. .41 .04 –.13 –.07 .12
*MQ 9: Usually when I have distressing thoughts or images, I “step back” and am aware of the thought or image without getting taken over by it. .44 .08 –.15 –.05 .14
*MQ 10: Usually when I have distressing thoughts or images, I just notice them and let them go. .44 –.15 –.16 –.03 .13

Factor 2: Observing/noticing/attending to sensations/perceptions/thoughts/feelings


FMI 3: I sense my body, whether eating, cooking, cleaning, or talking. .14 .50 –.03 –.03 –.03
FMI 6: I notice how my emotions express themselves through my body. .12 .44 .00 –.11 .02
FMI 7: I remain present with sensations and feelings even when they are unpleasant or painful. .14 .41 .00 –.07 .02
FMI 20: I examine pleasant as well as unpleasant sensations and perceptions. .18 .42 .07 –.16 .07
KIMS 1: I notice changes in my body, such as whether my breathing slows down or speeds up. –.11 .44 –.02 –.01 –.07
KIMS 5: I pay attention to whether my muscles are tense or relaxed. .05 .47 –.19 .05 –.08
*KIMS 9: When I’m walking, I deliberately notice the sensations of my body moving. –.02 .59 –.10 .00 –.11
*KIMS 13: When I take a shower or a bath, I stay alert to the sensations of water on my body. .02 .60 –.03 –.02 –.03
*KIMS 17: I notice how foods and drinks affect my thoughts, bodily sensations, and emotions. .06 .50 –.02 –.07 –.08
*KIMS 21: I pay attention to sensations, such as the wind in my hair or sun on my face. .03 .66 –.02 .02 –.01
*KIMS 25: I pay attention to sounds, such as clocks ticking, birds chirping, or cars passing. –.04 .60 .00 –.02 .02
*KIMS 29: I notice the smells and aromas of things. –.03 .56 .00 –.05 .10
KIMS 30: I intentionally stay aware of my feelings. –.01 .49 .00 –.22 –.08
*KIMS 33: I notice visual elements in art or nature, such as colors, shapes, textures, or patterns of light and shadow. .00 .52 .03 –.11 .01
*KIMS 37: I pay attention to how my emotions affect my thoughts and behavior. –.03 .50 –.08 –.13 –.27

Factor 3: Acting with awareness/automatic pilot/concentration/nondistraction


MAAS 2: I break or spill things because of carelessness, not paying attention, or thinking of something else. –.02 .02 –.48 –.02 .07
*MAAS 3: I find it difficult to stay focused on what’s happening in the present. .09 –.06 –.66 –.09 .03
*MAAS 7: It seems I am “running on automatic” without much awareness of what I’m doing. –.19 .27 –.66 .07 .07
*MAAS 8: I rush through activities without being really attentive to them. –.13 .24 –.67 .00 .13
MAAS 9: I get so focused on the goal I want to achieve that I lose touch with what I am doing right now to get there. –.16 .15 –.48 .10 .17
*MAAS 10: I do jobs or tasks automatically, without being aware of what I’m doing. –.12 .25 –.61 .14 .10
MAAS 11: I find myself listening to someone with one ear, doing something else at the same time. –.08 .16 –.45 .02 .08
MAAS 12: I drive places on “automatic pilot” and then wonder why I went there. –.22 .15 –.50 .06 .10
MAAS 13: I find myself preoccupied with the future or the past. .17 –.03 –.43 .04 .21
*MAAS 14: I find myself doing things without paying attention. –.09 .08 –.70 –.02 .11
MAAS 15: I snack without being aware that I’m eating. .05 .01 –.40 –.02 .13
FMI 9: I easily get lost in my thoughts and feelings. .17 –.21 –.54 –.02 .05
*KIMS 3: When I do things, my mind wanders off and I’m easily distracted. .10 –.14 –.64 –.14 –.12
KIMS 11: I drive on “automatic pilot” without paying attention to what I’m doing. –.20 .10 –.46 –.03 .11
*KIMS 23: I don’t pay attention to what I’m doing because I’m daydreaming, worrying, or otherwise distracted. .09 –.09 –.58 –.21 .03
KIMS 27: When I’m doing chores, such as cleaning or laundry, I tend to daydream or think of other things. .07 –.21 –.44 .02 –.06
KIMS 31: I tend to do several things at once rather than focusing on one thing at a time. –.05 .02 –.44 .00 .00
KIMS 35: When I’m working on something, part of my mind is occupied with other things, such as what I’ll be doing later or things I’d rather be doing. .14 –.14 –.55 –.06 –.09
CAMS 1: It is easy for me to concentrate on what I’m doing. .31 –.14 –.56 –.22 –.12
*CAMS 6: I am easily distracted. .18 –.17 –.65 –.19 –.14
CAMS 12: I am able to pay close attention to one thing for a long period of time. .30 –.11 –.51 –.23 –.11

Factor 4: Describing/labeling with words


*KIMS 2: I’m good at finding the words to describe my feelings. –.04 .05 .08 –.80 –.04
*KIMS 6: I can easily put my beliefs, opinions, and expectations into words. –.02 –.01 .00 –.76 .00
KIMS 10: I’m good at thinking of words to express my perceptions, such as how things taste, smell, or sound. –.05 .20 .00 –.65 .00
*KIMS 14: It’s hard for me to find the words to describe what I’m thinking. –.12 –.10 .00 –.80 .08
*KIMS 18: I have trouble thinking of the right words to express how I feel about things. –.19 –.09 .02 –.86 .12
*KIMS 22: When I have a sensation in my body, it’s hard for me to describe it because I can’t find the right words. –.15 –.02 –.03 –.65 .18
*KIMS 26: Even when I’m feeling terribly upset, I can find a way to put it into words. –.06 .08 .06 –.72 .03
*KIMS 34: My natural tendency is to put my experiences into words. –.11 .17 .05 –.71 –.06
*CAMS 5: I can usually describe how I feel at the moment in considerable detail. –.07 .06 .01 –.74 .00
CAMS 8: It’s easy for me to keep track of my thoughts and feelings. .08 .18 –.15 –.43 .00

Factor 5: Nonjudging of experience


*KIMS 4: I criticize myself for having irrational or inappropriate emotions. .13 –.09 –.16 –.07 .52
KIMS 8: I tend to evaluate whether my perceptions are right or wrong. –.15 –.17 .02 .11 .46
*KIMS 12: I tell myself that I shouldn’t be feeling the way I’m feeling. .09 –.09 –.10 –.12 .57
*KIMS 16: I believe some of my thoughts are abnormal or bad and I shouldn’t think that way. .00 –.08 .00 –.11 .67
*KIMS 20: I make judgments about whether my thoughts are good or bad. –.05 –.11 .00 .02 .69
KIMS 24: I tend to make judgments about how worthwhile or worthless my experiences are. .02 –.18 –.09 .02 .47
*KIMS 28: I tell myself I shouldn’t be thinking the way I’m thinking. .12 –.08 –.01 –.15 .67
*KIMS 32: I think some of my emotions are bad or inappropriate and I shouldn’t feel them. –.03 –.02 –.04 –.15 .73
*KIMS 36: I disapprove of myself when I have irrational ideas. .08 .00 .00 –.10 .64
MQ 6: Usually when I have distressing thoughts or images, I get angry that this happens to me. .21 .04 –.06 –.07 .41
*MQ 8: Usually when I have distressing thoughts or images, I judge myself as good or bad, depending what the thought/image is about. .11 .02 .11 –.05 .58

Initial eigenvalues for each factor 17.85 8.60 5.45 4.16 3.71
Percentage of variance accounted for after extraction 15.39 7.10 4.30 3.26 2.62

NOTE: MAAS = Mindful Attention Awareness Scale; FMI = Freiburg Mindfulness Inventory; KIMS = Kentucky Inventory of Mindfulness Skills; CAMS = Cognitive Affective Mindfulness Scale; MQ =
Mindfulness Questionnaire.
*Item included in five-factor mindfulness questionnaire in Parts III and V.

35
36 ASSESSMENT

TABLE 4 conducted for each facet in which all four of the remaining
Intercorrelations of Five Facets of Mindfulness facets were entered as predictors. The obtained value for
Derived From Exploratory Factor Analysis of adjusted R-squared represents the variance in each facet
Five Mindfulness Questionnaires accounted for by its relationship to the other four facets.
Facet Describe Actaware Nonjudge Nonreact Adjusted R2 values for the five facets ranged from .12 to
.23. Subtracting each facet’s R2 value from its alpha coeffi-
Observe .26** .15** –.07 .16** cient yields the systematic variance of the facet that is in-
Describe — .30** .21** .22**
Actaware — — .34** .33**
dependent of its relationship with the other four facets.
Nonjudge — — — .34** These values ranged from .56 to .75, showing that most of
the variance in each facet is distinct from the other four.
NOTE: Actaware = act with awareness. Overall, findings of Part 2 suggest that the combined
**p < .01.
item pool from the existing mindfulness measures in-
cludes five identifiable elements that are internally consis-
A total of 64 of the 112 items analyzed met our strict tent and only modestly correlated with each other. As can
criteria for inclusion in Table 3. For most of the remaining be seen in Table 3, four of the five questionnaires exam-
items, the highest factor loading was between .20 and .39, ined (all but the MAAS, which is unidimensional) contrib-
and many of these items showed similar loadings on more uted items to two or more of the five facets identified in the
than one factor. Thus, although many of these items proba- EFA, suggesting that most of the existing questionnaires
bly are reasonable representations of mindfulness, their include more than one facet. However, these findings do
content appears to incorporate more than one facet. As not clarify whether the five mindfulness facets derived
Smith et al. (2003) have noted, such items may be less use- from the EFA can be seen as elements of an overall mind-
ful than those loading on a single factor because they can fulness construct or are better understood as five separate
obscure the facet structure of the instrument, the relation- constructs. This question is examined in Part 3.
ships between facets and other measures, and the incre-
mental validity of some facets over others in predicting
relevant variables. PART 3: CONFIRMATORY FACTOR
ANALYSIS (CFA)
Deriving Mindfulness Facets
The purpose of Part 3 was to use CFA to investigate the
To create mindfulness facets with adequate internal replicability of the five-factor structure derived in Part 2 in
consistency and manageable length, the items with the an independent sample. Several models were tested, in-
highest loadings on the five factors derived in the EFA cluding hierarchical models that examine whether the five
were selected for inclusion in facet scales. For the non- facets should be viewed as components of an overarching
reactivity facet, all seven of the items shown in Table 3 mindfulness construct or are better understood as separate
were selected. For the other four facets (observing, acting constructs.
with awareness, nonjudging, describing), the eight items
with the highest factor loadings were selected, thus creat- Method
ing a subscale for each mindfulness facet. These items are
marked with asterisks in Table 3. Alpha coefficients for Participants and Procedures
each of the subscales then were computed. The following
alpha values were obtained: nonreactivity = .75, observ- A new sample of 268 undergraduate psychology stu-
ing = .83, acting with awareness = .87, describing = .91, dents (Sample 2) participated in Part 3. Their mean age
and nonjudging = .87. Thus, all five facet scales showed was 18.9 years, 77% were female, and 90% were Cauca-
adequate to good internal consistency. sian. Like Sample 1, they completed the procedures in ex-
Next, correlations among the five facet scales were change for research participation credit in their classes. In
computed. Because the facets appear to have distinct con- 1-hour sessions, they completed a brief demographic form
tent yet were all derived from questionnaires designed to and a mindfulness questionnaire created by combining the
measure mindfulness, correlations were expected to be 39 items (marked with asterisks in Table 3) that had been
modest but significant. As shown in Table 4, this pattern assigned to the five facets described in the preceding sec-
was found in most cases. Only one correlation was non- tion. For this instrument, called the Five Facet Mindful-
significant (observe with nonjudge). The others ranged ness Questionnaire (FFMQ), the 5-point Likert-type scale
from .15 to .34. from the KIMS was used (1 = never or very rarely true, 5 =
To provide additional evidence that the facets have sub- very often or always true). This required removing the
stantial nonoverlapping content, a regression analysis was word usually from the beginning of items from the MQ.
Baer et al. / FACETS OF MINDFULNESS 37

TABLE 5
Summary of Results of Tests of Alternative Factor Structures of Mindfulness
2 2
Model df c c difference CFI NNFI RMSEA

Sample 2 (n = 268)
One factor 90 1113.78*** — .43 .34 .21
Five factors 80 146.68*** — .96 .95 .06
Hierarchical—five factors 85 207.75*** 61.07* .93 .92 .07
Four factors 48 97.65*** — .97 .96 .06
Hierarchical—four factors 50 100.73*** 3.08 .97 .96 .06
Meditator sample (n = 190)
Five factors 80 154.44*** — .95 .94 .07
Hierarchical—five factors 82 149.69*** 3.01 .96 .95 .06

NOTE: Chi-square difference tests were conducted to compare each hierarchical model with its corresponding nonhierarchical model. The one-factor
model was ruled out; the difference tests were conducted to determine the value of a hierarchical framework. Hierarchical—five factors refers to the model
in which all five putative facets loaded on one, overall mindfulness factor. Hierarchical—four factors refers to the model in which describe, actaware,
nonjudge, and nonreact loaded on the overall mindfulness factor and observe was not included in the model. The meditator sample includes those partici-
pants from Samples 1 and 2 with meditation experience.
*p < .05. ***p < .001.

No other changes in the items were made. Items were ar- risk of spurious correlations is reduced, both because
ranged in an order that roughly alternated among the five fewer correlations are being estimated and because each
facets. Each participant also completed a subset of the cri- estimate is based on more stable indicators. Fourth, par-
terion measures from Part 1 to increase sample sizes for cels have been shown to provide more efficient estimates
correlational and regression analyses described later. of latent parameters than do items. Fifth, the object of in-
CFAs were conducted using the responses from these vestigation is not the performance of specific items but
268 participants to the 39 items on the FFMQ. In CFA, fit rather the relations among the scales. Before choosing to
indices indicate the extent to which the covariances among use item parcels, it is important to determine whether the
the items are accounted for by the hypothesized factor scale to be parceled represents a unidimensional construct;
model. We used four fit indices for these analyses: the if it does not, parcels of items could mask multidimension-
comparative fit index (CFI) and the nonnormed fit index ality (Hagtvet & Nasser, 2004; Little et al., 2002). Little
(NNFI; Bentler, 1990), the root mean square error of ap- et al. (2002) recommended conducting an exploratory fac-
proximation (RMSEA; Marsh, Balla, & Hau, 1996), and a tor analysis of the measure to evaluate scales’ unidimen-
chi-square test for discrepancy between the model and the sionality. We did so in Sample 1: Each of the five putative
data. By rule of thumb, CFI and NNFI values greater than facets of mindfulness emerged as unidimensional in that
.90 are thought to indicate good fit between a model and independent sample.
the data; for the RMSEA, a value of .05 is thought to indi- We took one further step to enhance our confidence in
cate close fit, .08 a fair fit, and .10 a marginal fit (Browne & these CFAs: We conducted all analyses twice, using meth-
Cudeck, 1993). Although the chi-square statistic is gener- ods recommended by Little et al. (2002). In the first set of
ally no longer used to evaluate fit because of its hypersen- analyses, we assigned items within scales to parcels ran-
sitivity, we report it here to facilitate comparisons between domly. In the second, items were assigned based on their
alternative factor models. We used the maximum likeli- factor loadings in the EFA. For both methods, we created
hood estimation method because of its robust performance three parcels for each factor and averaged the item scores
in a variety of situations (Hu, Bentler, & Kano, 1992). with each parcel. Thus, each a priori factor was repre-
For several reasons, we conducted these CFAs using sented by three indicators. A given factor’s indicators were
item parcels (groups of items) rather than individual items. not allowed to correlate with other factors, nor were error
Little, Cunningham, Shahar, and Widamon (2002) and terms allowed to correlate.
Rushton, Brainerd, and Pressley (1983) have described
several advantages of item parceling. First, the reliability Results
of a parcel of items is greater than that of a single item, so
parcels can serve as more stable indicators of a latent con- Results of tests of several alternative factor structures
struct. Second, as combinations of items, parcels provide are summarized in Table 5 for analyses using random par-
more scale points, thereby more closely approximating cel creation. Findings for the other parceling method
continuous measurement of the latent construct. Third, closely replicated these results. That is, model compari-
38 ASSESSMENT

sons produced the same results, and facet loadings for the these two scales. We examined this possibility by first
final model differed, on average, by only two one hun- combining Samples 1 and 2 (to increase the number of par-
dredths (.02). ticipants with meditation experience) and then comparing
First, we tested a single-factor model in which all item intercorrelations among the facets for participants with
parcels are indicators of one, overall mindfulness factor. and without meditation experience (ns = 190 and 667, re-
The fit of this model was poor—CFI = .43, NNFI = .34, spectively). Although most did not differ, the relation be-
and RMSEA = .21 (90% confidence interval: .20 to .22)— tween observe and nonjudge was significantly different
suggesting that the item parcels as a group do not have a (and positive) in those with meditation experience, sug-
unidimensional factor structure. Next, we tested the five- gesting that the observe facet might fit the hierarchical
factor model that was identified via EFA in the previous model described above in samples with more meditation
sample. The five factors were allowed to intercorrelate. experience.
This model fit the sample well: CFI = .96, NNFI = .95, and Testing this possibility in Sample 2 was not feasible be-
RMSEA = .06 (90% confidence interval: .04 to .07). This cause the number of participants with meditation experi-
finding replicates the results of the EFA on Sample 1. ence was too small for a CFA. Therefore, we tested this
However, it does not demonstrate whether the five factors model in those participants from our combined sample
are components of an overall mindfulness construct. To who reported some degree of meditation experience (n =
examine this question, we tested a hierarchical model, in 190) and found that all five facets loaded significantly on
which the five factors were themselves indicators of an the overall mindfulness construct (loadings: observe = .34,
overall mindfulness factor. This model fit the data reason- describe = .57, actaware = .72, nonjudge = .55, nonreact =
ably well: Both CFI and NNFI were greater than .90, and .71). Fit indices for this model were CFI = .96, NNFI = .94,
the RMSEA of .07 suggests a fair fit to the data. However, and RMSEA = .07, and the chi-square test of the difference
the pattern of loadings suggested that the model was mis- between this model and the five-factor nonhierarchical
specified. The loadings of describe, actaware, nonjudge, model in this sample was not significant, thus suggesting
and nonreact were all significant at p < .001, but observe the plausibility of a hierarchical, five-factor structure to
loaded nonsignificantly on the Overall Mindfulness fac- mindfulness among individuals with meditation experi-
tor. Not surprisingly, chi-square difference tests indicated ence. This model is depicted in Figure 2. This analysis
that this model fit significantly worse than did the five- must be interpreted cautiously because it included partici-
factor (nonhierarchical) model. pants from Sample 1 and therefore is not independent of
Next, we tested an alternative hierarchical model, in the EFA conducted in Part 2. However, it does suggest that
which describe, actaware, nonjudge, and nonreact were de- the hierarchical five-facet structure merits additional study
fined as facets of an overall mindfulness construct and ob- in an independent sample with a larger proportion of expe-
serve was not included in the model. That model fit much rienced meditators.
better (CFI = .97, NNFI = .96, RMSEA = .06). There was no
loss of fit for this more parsimonious four-factor hierarch-
ical model as compared to a four-factor nonhierarchical PART 4: DIFFERENTIAL RELATIONSHIPS
model (chi-square difference = 3.08, nonsignificant, all BETWEEN MINDFULNESS FACETS AND
fit indices identical). This finding supports a hierarchi- OTHER CONSTRUCTS
cal structure to mindfulness, in which describe, actaware,
nonjudge, and nonreact can be considered facets of a broad As noted earlier, measurement of complex constructs at
mindfulness construct. This model is depicted in Figure 1. the facet level is useful for understanding their relations
Failure of the observe facet to fit this model is unex- with other variables, particularly when one or more facets
pected, because observing is widely described as a central of a construct are strongly related to a specific variable,
feature of mindfulness. Lack of fit with the model is proba- whereas other facets are weakly related or unrelated.
bly a function of observe’s differential correlations with Therefore, the purpose of Part 4 was to examine whether
the other four facets, particularly its nonsignificant (and mindfulness facets are differentially related to the vari-
negative) correlation with nonjudge. In the development ables described in Part 1 (openness to experience, emo-
of the KIMS, Baer et al. (2004) reported a significant nega- tional intelligence, etc.). Based on the item content of the
tive correlation between observe and nonjudge and sug- mindfulness facets and the measures of the other con-
gested that in individuals with no meditation experience, structs, rational predictions were developed about which
attending to experiences might typically be associated mindfulness facet(s) should most strongly correlate with
with judging them but that people with meditation experi- each variable. Support for these predictions will provide
ence should be expected to show higher levels of both ob- evidence for the use of a multifaceted conceptualization of
serving and nonjudging and a positive correlation between mindfulness by showing which elements of mindfulness
Baer et al. / FACETS OF MINDFULNESS 39

FIGURE 1
The Hierarchical Model of Mindfulness for Sample 2

Mindfulness

.51
.39
.79 .64

describ
describe act NJ NR
e

.88 .83 .86 .90 .82 .82 .79 .82 .79 .64 .65 .61

P1 P2 P3 P1 P2 P3 P1 P2 P3 P1 P2 P3

NOTE: The coefficients describing the loadings of the four facets on the broad mindfulness construct are maximum likelihood estimates. Act = act with
awareness; NJ = nonjudging; NR = nonreactivity. Each of the four facets was represented by three parcels of items as indicators (P1 = Parcel 1, P2 = Parcel
2, etc.). Maximum likelihood estimates of parcel loadings on facets are also provided. For ease of presentation, error terms for parcels are omitted.

are responsible for the significant relationships shown in facets were expected to correlate with these constructs in
Part 1 between global mindfulness and these other con- the same directions as do the original mindfulness ques-
structs. Although the observe facet was not part of the hier- tionnaires (as shown in Table 2). That is, correlations
archical factor structure of mindfulness in our largely should be positive with constructs that appear to include
nonmeditating sample (see Part 3), it is included here be- elements of mindfulness (e.g., emotional intelligence,
cause of its potential use in samples of experienced self-compassion) and negative with constructs that appear
meditators. to reflect an absence of mindfulness (e.g., dissociation,
thought suppression, etc). Second, we made the following
Method predictions about which specific mindfulness facets
would be most strongly related to each variable. An impor-
Mindfulness facets were correlated with each of the tant feature of openness to experience is attentiveness to
other constructs. For these analyses, data from Samples 1 internal and external stimuli. Therefore, we predicted that
and 2 were combined. Total N for this combined sample openness would be most strongly related to the observe
was 881. However, as each participant had completed only facet. For both emotional intelligence and alexithymia,
a subset of the other measures, sample sizes for the corre- the ability to recognize and label emotional states is a cen-
lations ranged from 300 to 581. tral element. For this reason, these two variables were
We made the following predictions about relationships expected to be most strongly correlated with the describe
between mindfulness facets and other constructs. First, the facet. Dissociation and absent-mindedness both involve
40 ASSESSMENT

FIGURE 2
The Hierarchical Model of Mindfulness for the Sample With Meditation Experience

Mindfulness

.57
.72 .55
.71 .34

describ
describe act NJ NR observe
e

.85 .89 .93


.85 .89 .85 .83 .88
.86 .85 .67 .85 .73 .72
.68

P1 P2 P3 P1 P2 P3 P1 P2 P3 P1 P2 P3 P1 P2 P3

NOTE: Participants are from Samples 1 and 2. The coefficients describing the loadings of the five facets on the broad mindfulness construct are maximum
likelihood estimates. Act = act with awareness; NJ = nonjudging; NR = nonreactivity. Each of the five facets was represented by three parcels of items as in-
dicators (P1 = Parcel 1, P2 = Parcel 2, etc.). Maximum likelihood estimates of parcel loadings on facets are also provided. For ease of presentation, error
terms for parcels are omitted.

acting without awareness and therefore were expected to out judgment all were significantly and similarly
be most strongly related (negatively) to the act with aware- correlated with these variables, whereas observe was not.
ness facet, which measures attending to one’s current ac- Similar results were predicted here.
tivities and avoiding automatic pilot. Thought suppression
and difficulties in emotion regulation both include judg- Results and Discussion
mental or self-critical attitudes about thoughts and/or
emotions and therefore were expected to be most strongly Correlations between mindfulness facets and related
related to the nonjudging facet. Similarly, experiential variables can be seen in Table 6. Because of the large num-
avoidance and self-compassion both include judgmental ber of correlations presented, only those with p < .001 are
or self-critical attitudes. In addition, experiential avoid- marked as significant. In each row, the largest correlation
ance includes the ability to experience unpleasant inner is shown in bold, and the correlations that differ signifi-
phenomena without reacting to them with maladaptive, cantly (p < .01) from the largest one are shown in italics.
counterproductive behavior, and self-compassion in- These findings clearly show that the mindfulness facets
volves awareness of internal experience without maladap- are differentially related to the other constructs and that the
tive reactivity when suffering. Thus, experiential avoid- facets most strongly related to each construct are consis-
ance and self-compassion were expected to be strongly tent with our predictions. In each case, although several
and similarly correlated with both the nonjudging and facets are significantly correlated with the construct, the
nonreactivity facets. Finally, it was difficult to specify a predicted facet is significantly more strongly correlated
single mindfulness facet that should be most strongly re- than several others. In addition, the most strongly related
lated to psychological symptoms or neuroticism. Data col- facet differs across the constructs examined, suggesting
lected in the development of the KIMS (Baer et al., 2004) that all facets are useful in understanding the relationships
showed that describe, act with awareness, and accept with- between mindfulness and other conceptually related vari-
Baer et al. / FACETS OF MINDFULNESS 41

TABLE 6
Correlations Between Mindfulness Facets and Related Constructs
Mindfulness Facet
Construct Observe Describe Actaware Nonjudge Nonreact

Predicted positive correlations


Openness to experience .42*** .19*** .02 –.07 .18***
Emotional intelligence .22*** .60*** .31*** .37*** .21***
Self-compassion .14*** .30*** .40*** .48*** .53***
Predicted negative correlations
Alexithymia –.08 –.68*** –.42*** –.34*** –.19***
Dissociation .27*** –.32*** –.62*** –.49*** –.12
Absent-mindedness .16*** –.28*** –.61*** –.41*** –.15***
Psychological symptoms .17*** –.27*** –.48*** –.50*** –.31***
Neuroticism .07 –.23*** –.44*** –.55*** –.35***
Thought suppression .16*** –.23*** –.36*** –.56*** –.22***
Difficulties emotion regulation –.02 –.38*** –.40*** –.52*** –.36***
Experiential avoidance .12 –.23*** –.30*** –.49*** –.39***

NOTE: In each row, the largest correlation is shown in bold, and correlations that differ significantly from the largest (p < .01) are shown in italics.
***p < .001.

ables. For example, the describe facet is the most impor- TABLE 7
tant in understanding mindfulness’ relationships with Regression Analysis Showing Prediction of
emotional intelligence and alexithymia, whereas the act Psychological Symptoms by
with awareness facet is central to its relationships with dis- Mindfulness Facets
sociation and absent-mindedness. Variable B SE b t Significance
Directional findings are almost entirely consistent with
predictions, except for the observe facet, which was posi- Describe 3.21 .005 –.06 –1.47 .142
Actaware –.03 .005 –.29 –6.14 .000
tively correlated with openness, emotional intelligence, Nonjudge –.04 .005 –.36 –7.99 .000
and self-compassion (as predicted) and also positively Nonreact –.02 .006 –.11 –2.54 .012
correlated with dissociation, absent-mindedness, psycho- 2
logical symptoms, and thought suppression (contrary to NOTE: Actaware = act with awareness. R for model = .37.
our predictions). When recalculated in the subsample of
participants with meditation experience, these four unex- as measured by the BSI. Entry of more than one mindful-
pected positive correlations were nonsignificant, whereas ness facet into the model will suggest that consideration of
all other correlations in this table were unchanged or be- multiple facets is helpful in understanding the relationship
came significantly larger in the predicted direction. These between mindfulness and symptom level. Results of this
findings are consistent with those described earlier in sug- analysis can be seen in Table 7. Three of the four facets
gesting that observe’s relationships with other variables (actaware, nonjudge, nonreact) were significant predic-
may change as a function of meditation experience, tors, showing that each accounts for a significant portion
whereas this is apparently not true for the other four facets. of the variance not accounted for by the others. That is, it
appears that these three facets have incremental validity
over the others in the prediction of symptom level.
PART 5: INCREMENTAL VALIDITY OF
MINDFULNESS FACETS IN PREDICTING
PSYCHOLOGICAL SYMPTOMS GENERAL DISCUSSION

Most theoretical and empirical writings about mindful- The purposes of this project were to examine psycho-
ness address its use in reducing symptoms or improving metric characteristics of recently developed mindfulness
well-being. Thus, it is important to examine the extent to questionnaires, to use these instruments to investigate the
which mindfulness facets predict general mental health. facet structure of mindfulness, to examine whether identi-
For this purpose, we conducted a regression analysis in fied facets are differentially correlated with a variety of
which the mindfulness facets (excluding observe, which constructs that are conceptually related to mindfulness,
was correlated in the unexpected direction with symp- and to test whether facets have incremental validity in the
toms) were used to predict psychological symptom level prediction of psychological symptoms. Findings suggest
42 ASSESSMENT

several conclusions. First, the available mindfulness ques- the CAMS, and “I accept unpleasant experience” from the
tionnaires appear psychometrically promising, showing FMI). However, none of these items met our strict criteria
good internal consistency and expected correlations with for inclusion in Table 3, because they generally had mod-
several other variables. This was true even for the FMI, est and similar loadings on more than one factor. This find-
which was developed with experienced meditators and ing may be consistent with those of Brown and Ryan
whose authors have expressed concern about its use in (2004), who noted that the original form of the MAAS had
nonmeditating samples (Buchheld et al., 2001). an acceptance factor but that this factor showed no incre-
Findings also support the conceptualization of mind- mental validity in the prediction of criterion measures.
fulness as a multifaceted construct. Results of EFA sug- Items using the term accept may be less useful than other
gested that five distinct facets are represented within the items in clarifying the facets of mindfulness, perhaps be-
currently available mindfulness questionnaires. Correla- cause some respondents may equate acceptance with
tional analyses showed that four of these facets (describe, approval of undesirable conditions or with passive resig-
act with awareness, nonjudge, and nonreact) are consis- nation (Linehan, 1993a; Segal et al., 2002). However, our
tently related in expected ways to a variety of other vari- findings clearly suggest that nonreactivity and nonjudging
ables, whereas observe showed both expected and unex- of inner experience are useful facets. Both may be seen as
pected relationships. CFA suggested that describe, act ways of operationalizing acceptance. That is, to accept an
with awareness, nonjudge, and nonreact are elements of an experience, such as feeling anxious, might include refrain-
overarching mindfulness construct, and three of these fac- ing from judgments or self-criticism about having this ex-
ets (act with awareness, nonjudge, and nonreact) were perience (nonjudging) and refraining from impulsive reac-
shown to have incremental validity in the prediction of tions to the experience (nonreactivity). Additional work is
psychological symptoms. needed to clarify the definition and components of accep-
Findings for the observe facet were unexpected in two tance and its relationship to mindfulness.
ways. First, observe did not fit the hierarchical model in Several authors have noted the importance of discrimi-
our full CFA sample, although it fit well with a sample nating outcomes of practicing mindfulness from elements
having some exposure to meditation. In addition, ob- of the mindfulness construct. For example, Bishop et al.
serve’s correlations with a few of the other constructs were (2004) suggested that nonreactivity and compassion, al-
in the unexpected direction. Reasons for these findings are though sometimes discussed as components of mindful-
not entirely clear. It is possible that the content of the ob- ness, might be better understood as outcomes of mindful-
serve items used here does not adequately capture the ness practice, and Brown and Ryan (2004) made a similar
quality of noticing or attending to experience that is char- point about acceptance. The same question might be
acteristic of mindfulness. Several of the items included on raised about some of the facets of mindfulness identified in
the observe facet address external stimuli (sounds, smells, the current project. Although our data do not entirely re-
etc.) and bodily sensations, whereas the other facets are solve this question, our hierarchical CFAs and the finding
concerned primarily with cognitions and emotions or with of incremental validity of several facets in predicting psy-
functioning on automatic pilot. Perhaps observe items chological symptoms both suggest that the multifaceted
with similar content would show more of the expected pat- conceptualization of mindfulness has merit. In addition,
terns. It should also be noted that although mindfulness many current descriptions suggest that a mindful approach
has both statelike and traitlike qualities (Brown & Ryan, to experience includes, at a minimum, observing experi-
2004; Segal et al., 2004), it has also been described as a ences without reactivity or judgment and avoiding auto-
skill (or set of skills) that can be developed with practice matic pilot. On the other hand, confounding elements of
(Bishop et al., 2004; Linehan, 1993b). Therefore, it is pos- mindfulness with its outcomes will certainly impair our
sible that the observe facet is particularly sensitive to understanding of this important construct. Therefore, ad-
changes with meditation experience that alter its relation- ditional study of this question appears warranted.
ships with other mindfulness facets and with related vari- This project relied entirely on student samples. Several
ables, such that observe becomes a clear facet of mindful- authors have argued that mindfulness is a naturally occur-
ness and related in expected directions to other variables as ring characteristic that shows meaningful variation in non-
mindfulness skills develop. Additional work is required to clinical and nonmeditating samples (Brown & Ryan,
investigate this possibility, especially in samples with 2003; Kabat-Zinn, 2003). The numerous expected rela-
better representation of meditation experience. tionships reported here support this idea. However, be-
Our findings may shed light on the nature of accep- cause mindfulness-based interventions are used primarily
tance, which is often discussed as a central component of in clinical samples to address significant mental health is-
mindfulness. Several items using acceptance-related sues, the use of the questionnaires and the facet structure
terms are available in the item pool examined here (e.g., “I examined here must be investigated in clinical samples. In
am able to accept the thoughts and feelings I have” from addition, our findings suggest that the factor structure of
Baer et al. / FACETS OF MINDFULNESS 43

mindfulness and its relationships to other variables will be REFERENCES


enhanced by study of samples with meditation experience.
In spite of the reliance on student samples, these find- Baer, R. A. (2003). Mindfulness training as a clinical intervention: A
ings may have implications for clinical practice. Santorelli conceptual and empirical review. Clinical Psychology: Science and
Practice, 10, 125-143.
and Kabat-Zinn (2002) stated that the ability to describe
Baer, R. A., Smith, G. T., & Allen, K. B. (2004). Assessment of mindful-
mindfulness in readily accessible language is essential for ness by self-report: The Kentucky Inventory of Mindfulness Skills.
providers of MBSR. Understanding of empirically sup- Assessment, 11, 191-206.
ported facets of mindfulness may provide suggestions to Bagby, R. M., Taylor, G. J., & Parker, J. D. A. (1993). The twenty-item
Toronto Alexithymia Scale-II: Convergent, discriminant, and concur-
clinicians about how to describe it to clients. In addition, rent validity. Journal of Psychosomatic Research, 38, 33-40.
correlational and regression analyses showed that four of Bentler, P. M. (1990). Comparative fit indices in structural models. Psy-
the mindfulness facets are significantly related to psycho- chological Bulletin, 107, 238-246.
logical symptoms (in the expected direction) and that three Bishop, S. R., Lau, M., Shapiro, S., Carlson, L., Anderson, N. C.,
of these contribute independently to the prediction of Carmody, J., et al. (2004). Mindfulness: A proposed operational defi-
nition. Clinical Psychololgy: Science and Practice, 11, 230-241.
symptom level, suggesting that these skills are important Broadbent, B. E., Cooper. P. F., Fitzgerald, P., & Parkes, K. R. (1982). The
in teaching mindfulness for purposes of symptom reduc- cognitive failures questionnaire (CFQ) and its correlates. British
tion. Research on changes with treatment in these facets Journal of Clinical Psychology, 21, 1-16.
might shed additional light on how mindfulness training Brown, K. W., & Ryan, R. M. (2003). The benefits of being present:
Mindfulness and its role in psychological well-being. Journal of Per-
contributes to reductions in symptoms and improved well- sonality and Social Psychology, 84, 822-848.
being. Brown, K. W., & Ryan, R. M. (2004). Perils and promise in defining and
Conducting such research will require multifaceted as- measuring mindfulness: Observations from experience. Clinical Psy-
chology: Science and Practice, 11, 242-248.
sessment of mindfulness. The FFMQ derived here re-
Browne, M. W., & Cudeck, R. (1993). Alternative ways of assessing
quires extensive additional validation in a range of sam- model fit. In K. A. Bollen & J. S. Long (Eds.), Testing structural equa-
ples. However, our findings suggest that it has reasonable tion models (pp. 136-162). Newbury Park, CA: Sage.
psychometric properties and it is currently the only one Buchheld, N., Grossman, P., & Walach, H. (2001). Measuring mindful-
ness in insight meditation (Vipassana) and meditation-based psycho-
that assesses all five of the mindfulness facets identified in therapy: The development of the Freiburg Mindfulness Inventory
this project. Thus, it may be useful in future research. (FMI). Journal for Meditation and Meditation Research, 1, 11-34.
An alternative for multifaceted assessment is the KIMS, Chadwick, P., Hember, M., Mead, S., Lilley, B., & Dagnan, D. (2005). Re-
which measures four of the five facets identified here (all sponding mindfully to unpleasant thoughts and images: Reliability
and validity of the Mindfulness Questionnaire. Unpublished
but nonreactivity). manuscript.
Future research should expand the assessment of mind- Cioffi, D., & Holloway, J. (1993). Delayed costs of suppressed pain. Jour-
fulness to include methods other than self-report question- nal of Personality and Social Psychology, 64, 274-282.
naires. Experience sampling methods could be used to Clark, L. A., & Watson, D. (1995). Constructing validity: Basic issues in
objective scale development. Psychological Assessment, 7, 309-319.
examine individuals’ mindful awareness during daily ac-
Costa, P. T., & McCrae, R. R. (1992). Revised NEO Personality Inventory
tivities (Brown & Ryan, 2003). In addition, a growing (NEO PI-R) and NEO Five-Factor Inventory (NEO-FFI): Profes-
literature uses laboratory tasks to study the effects of ac- sional manual. Odessa, FL: Psychological Assessment Resources.
ceptance and suppression-based methods of coping with Derogatis, L. R. (1992). The Brief Symptom Inventory (BSI): Ad-
minstration, scoring and procedures manual=II (2nd ed.). Minneap-
stressors, such as upsetting film clips, cold-pressor task, or olis, MN: National Computer Systems.
carbon dioxide challenge. Findings generally suggest that Dimidjian, S., & Linehan, M. M. (2003a). Defining an agenda for future
acceptance-based methods lead to increased tolerance of research on the clinical application of mindfulness practice. Clinical
stressful stimuli (e.g., Cioffi & Holloway, 1993; Feldner, Psychology: Science and Practice, 10, 166-171.
Dimidjian, S., & Linehan, M. M. (2003b). Mindfulness practice. In
Zvolensky, Eifert, & Spira, 2003; S. C. Hayes, Bissett, et W. O’Donohue, J. E. Fisher, & S. C. Hayes (Eds.), Empirically sup-
al., 1999; Levitt, Brown, Orsillo, & Barlow, 2004). Addi- ported techniques of cognitive behavior therapy: A step-by-step
tional studies could examine whether differences in the guide for clinicians. New York: John Wiley.
mindfulness facets identified here are related to differ- Feldman, G. C., Hayes, A. M., Kumar, S. M., & Greeson, J. M. (2004).
Development, factor structure, and initial validation of the Cognitive
ences in tendencies to use mindfulness-based strategies and Affective Mindfulness Scale. Unpublished manuscript.
for coping with such laboratory stressors and whether Feldner, M. T., Zvolensky, M. J., Eifert, G. H., & Spira, A. P. (2003). Emo-
training in these mindfulness skills increases tolerance for tional avoidance: an experimental test of individual differences and
them. response suppression using biological challenge. Behaviour Re-
search and Therapy, 41, 403-411.
In summary, findings reported here suggest that self- Fischer, S., Smith, G. T., & Anderson, K. G. (2003). Clarifying the role of
report mindfulness questionnaires have good psycho- impulsivity in bulimia nervosa. International Journal of Eating Dis-
metric properties and that exploration of the facets of orders, 33, 406-411.
mindfulness may be useful in understanding the nature of Fischer, S. Smith, G. T., Anderson, K. G., & Flory, K. (2003). Expectan-
cies influence the operation of personality on behavior. Psychology of
the construct and its relationships with other variables. Addictive Behaviors, 17, 108-114.
44 ASSESSMENT

Floyd, F. J., & Widamon, K. F. (1995). Factor analysis in the development Marlatt, G. A., & Gordon, J. R. (1985). Relapse prevention: Mainte-
and refinement of clinical assessment instruments. Psychological nance strategies in the treatment of addictive behaviors. New York:
Assessment, 3, 286-299. Guilford.
Goldstein, J. (2002). One Dharma: The emerging Western Buddhism. San Marlatt, G. A., & Kristeller, J. L. (1999). Mindfulness and meditation.
Francisco: Harper Collins. In W. R. Miller (Ed.), Integrating spirituality into treatment (pp. 67-
Gratz, K. L., & Roemer, L. (2004). Multidimensional assessment of emo- 84). Washington, DC: American Psychological Association.
tion regulation and dysregulation: Development, factor structure, and Marsh, H. W., Balla, J. R., & Hau, K. Y. (1996). An evaluation of in-
initial validation of the difficulties in emotion regulation scale. Jour- cremental fit indices: A clarification of mathematical and empiri-
nal of Psychopathology and Behavioral Assessment, 26, 41-54. cal properties. In G. A. Marcoulides & R. E. Schumacker (Eds.),
Hagtvet, K. A., & Nasser, F. M. (2004). How well do item parcels repre- Advanced structural equation modeling: Issues and techniques
sent conceptually defined latent constructs? A two-facet approach. (pp. 315-353). Mahwah, NJ: Lawrence Erlbaum.
Structural Equation Modeling, 11, 168-193. Martinez-Pons, M. (1997). The relation of emotional intelligence with
Hayes, A. M., & Feldman, G. (2004). Clarifying the construct of mindful- selected areas of personal functioning. Imagination, Cognition, and
ness in the context of emotion regulation and the process of change in Personality, 17, 3-13.
therapy. Clinical Psychology: Science and Practice, 11, 255-262. Mayer, J. L., & Farmer, R. F. (2003). The development and psychometric
Hayes, A. M., & Harris, M. S. (2000). The development of an integrative evaluation of a new measure of dissociative activities. Journal of Per-
treatment for depression. In S. Johnson, A. M. Hayes, T. Field, sonality Assessment, 80, 185-196.
N. Schneiderman, & P. McCabe (Eds), Stress, coping, and depression Neff, K. D. (2003a). The development and validation of a scale to mea-
(pp. 291-306). Mahwah, NJ: Lawrence Erlbaum. sure self-compassion. Self and Identity, 2, 223-250.
Hayes, S. C., Bissett, R. T., Korn, Z., Zettle, R. D., Rosenfarb, I. S., Neff, K. D. (2003b). Self-compassion: An alternative conceptualization
Cooper, L. D., et al. (1999). The impact of acceptance versus control of a healthy attitude toward oneself. Self and Identity, 2, 85-102.
rationales on pain tolerance. Psychological Record, 49, 33-47. Parks, G. A., Anderson, B. K., & Marlatt, G. A. (2001). Relapse preven-
Hayes, S. C., Masuda, A., Bissett, R., Luoma, J., & Guerrero, L. F. (2004). tion therapy. In N. Heather, T. J. Peters, & T. Stockwell (Eds.), Inter-
DBT, FAP, and ACT: How empirically oriented are the new behavior national handbook of alcohol dependence and problems (pp. 575-
therapy technologies? Behavior Therapy, 35, 35-54. 592). New York: John Wiley.
Hayes, S. C., Strosahl, K., & Wilson, K. G. (1999). Acceptance and com- Paunonen, S. V., & Ashton, M. C. (2001). Big Five factors and facets and
mitment therapy: An experiential approach to behavior change. New the prediction of behavior. Journal of Personality and Social Psychol-
York: Guilford. ogy, 81, 524-539.
Hayes, S. C., Strosahl, K., Wilson, K. G., Bissett, R. T., Batten, S. V., Robins, C. J., & Chapman, A. L. (2004). Dialectical behavior therapy:
Bergan, J., et al. (in press). The Acceptance and Action Question- Current status, recent developments, and future directions. Journal of
naire (AAQ) as a measure of experiential avoidance. Psychological Personality Disorders, 18, 73-89.
Record. Rushton, J. P., Brainerd, C. J., & Pressley, M. (1983). Behavioral develop-
Hayes, S. C., Wilson, K. D., Gifford, E. V., Follette, V. M., & Strosahl, K. ment and construct validity: The principle of aggregation. Psycholog-
(1996). Emotional avoidance and behavioral disorders: A functional ical Bulletin, 94, 18-38.
dimensional approach to diagnosis and treatment. Journal of Consult- Salovey, P., Mayer, J. D., Goldman, S. L., Turvey, C., & Palfai, T. P.
ing and Clinical Psychology, 64, 1152-1168. (1995). Emotional attention, clarity, and repair: Exploring emotional
Haynes, S. N., & Lench, H. C. (2003). Incremental validity of new clini- intelligence using the Trait Meta-Mood Scale. In J. W. Pennebaker
cal assessment measures. Psychological Assessment, 15, 456-466. (Ed.), Emotion, disclosure, and health (pp. 125-154). Washington,
Hough, L. M., & Schneider, R. J. (1995). Personality traits, taxonomies, and DC: American Psychological Association.
applications in orgainzations. In K. R. Murhpy (Ed.), Individuals and Santorelli, S., & Kabat-Zinn, J. (2002). Mindfulness-based stress re-
behavior in organizations (pp. 31-88). San Francisco: Jossey-Bass. duction professional training: MBSR curriculum guide and sup-
Hu, L. T., Bentler, P. M., & Kano, Y. (1992). Can test statistics in co- porting materials. Worcester: University of Massachusetts Medi-
variance structure analysis be trusted? Psychological Bulletin, 112, cal School.
351-362. Schneider, R. J., Hough, L. M., & Dunnette, M. D. (1996). Broadsided by
Kabat-Zinn, J. (1982). An outpatient program in behavioral medicine for broad traits: How to sink science in five dimensions or less. Journal of
chronic pain patients based on the practice of mindfulness medita- Organization Behavior, 17, 639-655.
tion: Theoretical considerations and preliminary results. General Segal, Z. V., Williams, J. M. G., & Teasdale, J. D. (2002). Mindfulness-
Hospital Psychiatry, 4, 33-47. based cognitive therapy for depression: A new approach to prevent-
Kabat-Zinn, J. (1990). Full catastrophe living: Using the wisdom of your ing relapse. New York: Guilford.
mind and body to face stress, pain, and illness. New York: Delacorte. Smith, G. T., Fischer, S., & Fister, S. M. (2003). Incremental validity prin-
Kabat-Zinn, J. (2000). Indra’s net at work: The mainstreaming of Dharma ciples in test construction. Psychological Assessment, 15, 467-477.
practice in society. In G. Watson & S. Batchelor (Eds.), The psychol- Smith, G. T., & McCarthy, D. M. (1995). Methodological considerations
ogy of awakening: Buddhism, science, and our day-to-day lives in the refinement of clinical assessment instruments. Psychological
(pp. 225-249). North Beach, ME: Weiser. Assessment, 7, 300-308.
Kabat-Zinn, J. (2003). Mindfulness-based interventions in context: Past, Wallace, J. C., Kass, S. J., & Stanny, C. J. (2002). The cognitive failures
present, and future. Clinical Psychology: Science and Practice, 10, questionnaire revisited: Dimensions and correlates. Journal of Gen-
144-156. eral Psychology, 129, 238-256.
Levitt, J. T., Brown, T. A., Orsillo, S. M., & Barlow, D. H. (2004). The Wegner, D. M., & Zanakos, S. (1994). Chronic thought suppression.
effects of acceptance versus suppression of emotion on subjective and Journal of Personality, 62, 615-640.
psychophysiological response to carbon dioxide challenge in patients Wenzlaff, R. M., & Wegner, D. M. (2000). Thought suppression. Annual
with panic disorder. Behavior Therapy, 35, 747-766. Review of Psychology, 51I, 59-91.
Linehan, M. M. (1993a). Cognitive-behavioral treatment of borderline
personality disorder. New York: Guilford.
Linehan, M. M. (1993b). Skills training manual for treating borderline Ruth A. Baer, PhD, is an associate professor of psychology at
personality disorder. New York: Guilford. the University of Kentucky. Her research interests include mind-
Little, T. D., Cunningham, W. A., Shahar, G., & Widamon, K. F. (2002). fulness and acceptance-based interventions, assessment and con-
To parcel or not to parcel: Exploring the question, weighing the mer- ceptualization of mindfulness, cognitive-behavioral interven-
its. Structural Equation Modeling, 9, 151-173.
tions, and psychological assessment.
Baer et al. / FACETS OF MINDFULNESS 45

Gregory T. Smith, PhD, is a professor of psychology and direc- Jennifer Krietemeyer, BA, is a graduate student in the clini-
tor of clinical training at the University of Kentucky. His research cal psychology program at the University of Kentucky. Her
interests include psychometric and validity theory, risk for eating research interests include mindfulness and acceptance-based
disorders and problem drinking, and integrating personality and interventions, eating disorders and obesity, and psychological
psychosocial models of risk. assessment.

Jaclyn Hopkins, BA, is a recent graduate with honors in psy- Leslie Toney, BA, is a graduate student in the clinical psychol-
chology from the University of Kentucky. Her research interests ogy program at the University of Kentucky. Her research inter-
include mindfulness and acceptance-based interventions, medi- ests include health psychology and multicultural psychology.
tation, health psychology, and psychological assessment.